5D10 + LEFT plugin working in 5441 (5440,5403)
Hello TekScopes! Thank you all for aiding and abetting. Here's a tiny bit of giving back.
The 5D10 Waveform Digitizer plugin, essentially a 1MS/s DSO, gets some credit for keeping the 5000 series oscopes interesting. While a 5D10 can digitize one or two channels from another plugin, only the "slow" 51xx scopes supported that capability. 5000-series docent Christian Weagle tried to adapt a "fast" 5440 with initial success that didn't pan out. I fell down this rabbit hole and eventually climbed out with a good result for my 5441. The same should work for a 5440 also - i.e. 5403 mainframes.
To enable LEFT plugin access for a 5D10 in a 5403/5440/5441 scope:
add three connections between plugin connectors on the interface board.
1. left A13 <---> center A17
2. left B13 <---> center A16
3. horizontal A19 <---> adjacent DIP14 7473 pin 6
The two left <---> center connections give the 5D10 access to the differential signal from the left plugin bay. That alone allows acquisition of a _single_ LEFT channel when the left plugin DISPLAY switch is _OFF_. A pair of 10cm F-F "dupont" jumpers made this connection trivially and reversibly.
The horizontal A19 <---> 7473 pin 6 connection lets the 5D10 keep control of the display when the left plugin DISPLAY switch is ON. Then the 5D10 can cycle the channel select logic to sample both channels from a dual channel plugin without chopping up the display.
Mr. Weagle offered a "reasonable guess" that Tek didn't support this configuration because adding extra load to the signal path would blunt high frequency response of the left plugin bay in normal operation of the faster 54xx scopes and later attributed disappointing results to hitting that problem hard. After much pondering how to deal with that, I eventually got around to trying it to assess the magnitude of the problem and found… no problem. By not-very-careful measurement, the added load attenuates a 6div 50MHz signal by a fraction of a trace width at most and a 100MHz signal by approximately not much. The mild loss of far-above-spec bandwidth from the left bay is good to know but I won't worry about it.
Beware. While the added jumpers themselves didn't bog down a 54xx-speed left plugin, Tek cautions that a horizontal plugin in the center bay certainly will. The manual says "nondestructively". The manual also only contemplates plugins compatible with 5100 series scopes. I haven't tried to work out if any of the faster plugins could be more vulnerable.
The procedure for modifying elder 51xx mainframes advises adding some resistors. 54xx scopes already have them.
Schematics for 5116 and 5111A (after the first 206 units) show ferrite beads on the left <---> center connections. Considering that no ferrites are advised for updating older scopes and Tek apparently shipped a couple hundred 5111As without them, I suppose that their potential benefit is less than compelling.
I have an Option 1 scope without the readout board. That allowed easy access to the back of the plugin interface board. For scopes with readout, removing the readout board may be helpful or necessary. The manual for installing the Option 3 custom readout wiring describes how to pull the readout board from a desktop scope by partially separating the back ends of the display and mainframe modules.
Doing this with a 5444 might be as simple as adding the two signal jumpers and turning the left beam intensity down. ?.
Working this out involved some side trips that might become future posts, but that will be for another day.
Please tell us if you verify this working in another 54xx scope.
 Ironically lauded for superior slowness
https://groups.io/g/TekScopes/message/82688 but 3 traces vs "dual trace"
(And the très spécial 5116+5D10 "Three Color, High Resolution Digital Storage Display"! http://w140.com/tekwiki/images/0/0a/5D10tb.jpg but that's its own thing)
 Older 51xx scopes require modification. See pages 4-21,23 of the service manual (ToC is rwong). This provides a jumping-off point for working out what to do with 54xx scopes. Manuals available here: http://w140.com/tekwiki/wiki/5D10
 AKA tertiary_42, i9t.net
Mr. Weagle stopped posting without comment after April 2010. Does anyone here know of his well-being?
 Aargh with naming plugin bays. "Left" & "horizontal" are unambiguous if not orthogonal. But "right vertical" vs "center" vs "left half of 5D10" per context complicate naming the one between the other two. "Center" clashes with Tek's left/right/horizontal nomenclature but depends less on context.
 May or may not work with DISPLAY=ON depending on indeterminate state of the channel switch logic. This may be part of why Mr. Weagle reported initial success then not.
 Like this:
The plugin connector pins are larger than 25mil so "Dupont" F pins need a firm push to fit over them. They might not fit securely on 25mil pins anymore but I don't expect to care. Fitting them with clearance under a readout board may require a little attention.
 The 7473 is dual flip-flop IC U780 in the interface board schematic. One ff toggles which channel dual channel plugins send to the mainframe and also clocks the other ff to alternately send left or "right vertical"/center plugin signals to the display when both vertical plugins set DISPLAY=ON (which the 5D10 does always). The schematic doesn't show unused pin 6 which is the CLR pin (CLeaR i.e. asynchronously RESET) for the plugin select ff. The 5D10 pulls horizontal plugin pin A19 low which in later 51xx mainframes asserts the asynchronous PRESET input of the 74113 type plugin select ff used in those instruments. While the 7473 ffs have no PRESET input, the 5403 logic is arranged differently so that asserting CLR does the right thing.
 To get some idea where trouble starts, I first clipped a pair of ~1foot 24awg silicone insulated leads to the A/B13 pins at the back of the interface board. Twisting them together cut a 50MHz signal by about 10% amplitude and 100MHz by about 50%, both roughly estimated. Laying the leads together without twisting had discernible but small effect at 50MHz and more noticeable but still small effect at 100MHz, no quantitative estimate attempted. So don't use foot-long jumpers.
Made more widely known by Christian Weagle: "The instructions give a good procedure to get [the readout board] out. I'd recommend following their advice about raising the rear of the display, by the way."